Buckling and growth of disbonds in honeycomb sandwich structure

Abstract

The response of honeycomb sandwich structure to disbond damage, in a compressive stress field, has been assessed. Two types of disbond were considered; those extending through the full width of a panel (through-width) and those enclosed within the panel boundaries (embedded). For each type of disbond the failure process was established through four-point bend testing of 56 sandwich specimens. For both types of disbond, failure was found to be governed by buckling-driven disbond growth and, consequently, models of buckling and disbond growth were developed. All of the models decoupled the face-sheets of the sandwich and treated the core as a spring foundation having a stiffness determined from equilibrium of a two-dimensional orthotropic solid. A linear Winkler beam model was used to predict buckling of a through-width disbond. The model buckling loads agreed with specimen test results with an average difference of 1.7%. A non-linear Winkler beam model was then developed to predict post-buckling behaviour and the initiation of disbond growth, through a fracture mechanics analysis. A characteristic growth curve, defining the work input required to initiate disbond growth, was developed and agreed with specimen test results with an average difference of 3.3%. The model also verified that disbond growth occurs in discrete increments approximately equal to the diameter of the honeycomb cells. A linear Winkler plate model was used to predict buckling of an embedded disbond. The model buckling loads agreed with specimen test results with an average difference of 3.7%. A non-linear Winkler plate model was then developed to predict post-buckling behaviour of a sandwich panel containing an embedded disbond. The model considered contact conditions and modelled disbond growth by releasing fractured nodes during load incrementation. Disbond growth initiation loads agreed with specimen test results with an average difference of 15.8%. Failure loads consistently over-predicted specimen test results by an average of 13.9%. It was concluded that the growth initiation loads should be used as a conservative estimate of failure. The models developed may be used to assess the criticality of disbond damage in sandwich structure having thin-gauge, composite face-sheets.